Timo Soeterik

124 CHAPTER 7 Predictors and outcome definitions Patient-based PSA and prostate volume, necessary for calculation of PSAD, were based on the most recent available measurements preoperatively. Prostate volume was measured by transrectal ultrasonography or mpMRI. Side-specific DRE staging information was collected before biopsy by the treating urologist during routine clinical care. Side-specific DRE and mpMRI staging information were both subdivided in three subclasses. These included non-palpable disease (T1), organ-confined nodal disease (T2) and EPE (T3) for DRE. As for mpMRI, these included non-visible lesions (T1), organ-confined lesions (T2) and lesions with EPE (T3). Imaging features used to assess EPE included thickening or suspicion for invasion of the neurovascular bundle, bulging of the prostatic contour, capsule irregularity, obliteration of the recto-prostatic angle, presence of a hypo-intensive signal in a periprostatic area and length of tumour contact with the capsule. 17-19 Explicit statements about presence or absence of EPE in the radiological report were scored accordingly. In less explicit cases a strong suspicion of EPE was classified as positive. Cases in which EPE could not be ruled out were classified as negative. 19 Side-specific biopsy information, including highest percentage of positive cores on systematic biopsy and highest ISUP grade, were documented during routine clinical care for both right and left lobe separately. Final surgical histopathological information including pathological tumour stage and highest ISUP grade found in the resected prostate specimen was documented on a whole-gland level. If EPE was observed, the laterality (left, right or both lobes) was reported. EPE was defined as a tumour that bulges beyond the prostate contour, tumour that is admixed with periprostatic fat tissue or, in the posterolateral area, as tumour within connective tissue or between nerves of the neurovascular bundle. EPE was distinct from microscopic bladder neck invasion (presence of tumour between thick smooth muscle bundles in the absence of benign prostate glands) and seminal vesicle invasion, which were not considered as EPE in our study. 20 The RP specimens were processed with conventional sections in 1810 (97%) cases and using whole-mount sections in 60 (3%) cases. Model building Four models were built according to the “full model” principle, 21 including combinations of five predictors corresponding with different staging work-up strategies. Model 1 consisted of PSAD, DRE, biopsy ISUP grade and percentage positive cores on systematic biopsy. Model 2 included PSAD, mpMRI and biopsy ISUP grade. Model 3 included PSAD, mpMRI, DRE and biopsy ISUP grade. Model 4 included all five predictors. For analysis purposes, the right and left prostatic lobe of each patient were regarded as separate cases. That is, for calculating the probability of EPE in the right lobe: patient-based PSAD, right- sided biopsy information, right-sided DRE staging information and right-sided mpMRI staging information were used.